{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T16:06:31Z","timestamp":1789401991447,"version":"build-2803163510"},"reference-count":49,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T00:00:00Z","timestamp":1789344000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Neurol."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Stroke-associated pneumonia (SAP) is a frequent complication after acute ischemic stroke (AIS) and is associated with poor outcomes. This study aimed to develop an interpretable multimodal deep learning model integrating MRI, lesion-related brain regions, and clinical variables for early SAP prediction.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>A total of 426 AIS patients were retrospectively enrolled, including 71 patients with SAP. Multimodal MRI data (DWI, T1WI, and T2-FLAIR) were processed using standardized registration and lesion segmentation. A 3D convolutional neural network was used to extract imaging representations, which were fused with clinical variables and AAL3-based brain-region features. Model performance was assessed using stratified five-fold cross-validation and nested cross-validation when applicable, with further evaluation based on receiver operating characteristic (ROC) analysis, calibration analysis, and decision curve analysis. Grad-CAM was applied for model interpretation.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The multimodal fusion model achieved the best performance for SAP prediction, with an AUC of 0.782 (95% CI: 0.712\u20130.839), compared with the clinical model based on conventional clinical variables (AUC\u202f=\u202f0.756), the imaging model based on 3D CNN representations (AUC\u202f=\u202f0.693), and the brain-region model based on AAL3-derived lesion location features (AUC\u202f=\u202f0.501). The fusion model showed superior clinical utility and favorable calibration. Grad-CAM visualization demonstrated that model predictions were mainly driven by lesion-related cortical and subcortical regions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>A multimodal deep learning framework integrating MRI, brain-region information, and clinical characteristics improved SAP prediction after AIS and provided an interpretable approach for individualized risk stratification.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fneur.2026.1917270","type":"journal-article","created":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T15:36:26Z","timestamp":1789400186000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Multimodal deep learning outperforms clinical and brain region models in predicting stroke-associated pneumonia: an explainable AI study"],"prefix":"10.3389","volume":"17","author":[{"given":"Haoyang","family":"He","sequence":"first","affiliation":[{"name":"Department of Neurosurgery, Renmin Hospital of Wuhan University","place":["Wuhan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xu","family":"Zhang","sequence":"additional","affiliation":[{"name":"Department of Neurosurgery, Renmin Hospital of Wuhan University","place":["Wuhan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lijuan","family":"Gu","sequence":"additional","affiliation":[{"name":"Central Laboratory, Renmin Hospital of Wuhan University","place":["Wuhan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhihong","family":"Jian","sequence":"additional","affiliation":[{"name":"Department of Neurosurgery, Renmin Hospital of Wuhan University","place":["Wuhan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaoxing","family":"Xiong","sequence":"additional","affiliation":[{"name":"Department of Neurosurgery, Renmin Hospital of Wuhan University","place":["Wuhan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,14]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"795","DOI":"10.1016\/S1474-4422(21)00252-0","article-title":"Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the global burden of disease study 2019","volume":"20","year":"1990","journal-title":"Lancet Neurol"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"107","DOI":"10.1186\/s12883-016-0627-y","article-title":"Post-stroke pneumonia at the stroke unit - a registry based analysis of contributing and protective factors","volume":"16","author":"Matz","year":"2016","journal-title":"BMC Neurol"},{"key":"ref3","doi-asserted-by":"publisher","first-page":"796","DOI":"10.1038\/nm.2399","article-title":"The immunology of stroke: from mechanisms to translation","volume":"17","author":"Iadecola","year":"2011","journal-title":"Nat Med"},{"key":"ref4","doi-asserted-by":"publisher","first-page":"110","DOI":"10.1186\/1471-2377-11-110","article-title":"Post-stroke infection: a systematic review and meta-analysis","volume":"11","author":"Westendorp","year":"2011","journal-title":"BMC Neurol"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"430","DOI":"10.1159\/000350199","article-title":"Stroke-associated pneumonia: major advances and obstacles","volume":"35","author":"Hannawi","year":"2013","journal-title":"Cerebrovasc Dis"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"293","DOI":"10.1111\/ane.12956","article-title":"Impact of stroke-associated pneumonia on mortality, length of hospitalization, and functional outcome","volume":"138","author":"Teh","year":"2018","journal-title":"Acta Neurol Scand"},{"key":"ref7","doi-asserted-by":"publisher","first-page":"401","DOI":"10.1038\/nrneurol.2012.98","article-title":"The immunology of acute stroke","volume":"8","author":"Chamorro","year":"2012","journal-title":"Nat Rev Neurol"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"3671","DOI":"10.1177\/0271678X16671964","article-title":"Stroke-induced immunodepression and dysphagia independently predict stroke-associated pneumonia - the PREDICT study","volume":"37","author":"Hoffmann","year":"2017","journal-title":"J Cereb Blood Flow Metab"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"1009041","DOI":"10.3389\/fnut.2022.1009041","article-title":"A nomogram based on nutritional status and a(2)DS(2) score for predicting stroke-associated pneumonia in acute ischemic stroke patients with type 2 diabetes mellitus: a retrospective study","volume":"9","author":"Song","year":"2022","journal-title":"Front Nutr"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"2619","DOI":"10.1016\/j.jstrokecerebrovasdis.2015.07.017","article-title":"External validation of the Prestroke Independence, sex, age, National Institutes of Health stroke scale (ISAN) score for predicting stroke-associated pneumonia in the Athens stroke registry","volume":"24","author":"Papavasileiou","year":"2015","journal-title":"J Stroke Cerebrovasc Dis"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"198","DOI":"10.1136\/jnnp.2009.182394","article-title":"The impact of lesion location and lesion size on poststroke infection frequency","volume":"81","author":"Minnerup","year":"2010","journal-title":"J Neurol Neurosurg Psychiatry"},{"key":"ref12","doi-asserted-by":"publisher","first-page":"441","DOI":"10.1016\/j.ejca.2011.11.036","article-title":"Radiomics: extracting more information from medical images using advanced feature analysis","volume":"48","author":"Lambin","year":"2012","journal-title":"Eur J Cancer"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"563","DOI":"10.1148\/radiol.2015151169","article-title":"Radiomics: images are more than pictures, they are data","volume":"278","author":"Gillies","year":"2016","journal-title":"Radiology"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"500","DOI":"10.1038\/s41568-018-0016-5","article-title":"Artificial intelligence in radiology","volume":"18","author":"Hosny","year":"2018","journal-title":"Nat Rev Cancer"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"1030","DOI":"10.1044\/2019_AJSLP-19-00068","article-title":"Cortical and subcortical control of swallowing-can we use information from lesion locations to improve diagnosis and treatment for patients with stroke?","volume":"29","author":"Wilmskoetter","year":"2020","journal-title":"Am J Speech Lang Pathol"},{"key":"ref16","doi-asserted-by":"publisher","first-page":"904","DOI":"10.1212\/01.wnl.0000269789.09277.47","article-title":"Statin treatment withdrawal in ischemic stroke: a controlled randomized study","volume":"69","author":"Blanco","year":"2007","journal-title":"Neurology"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"1438","DOI":"10.1161\/STROKEAHA.122.038867","article-title":"Immunodepression, infections, and functional outcome in ischemic stroke","volume":"53","author":"Westendorp","year":"2022","journal-title":"Stroke"},{"key":"ref18","doi-asserted-by":"publisher","first-page":"190","DOI":"10.1186\/s12931-025-03272-z","article-title":"Decoding mechanisms and protein markers in lung-brain axis","volume":"26","author":"Huang","year":"2025","journal-title":"Respir Res"},{"key":"ref19","doi-asserted-by":"publisher","first-page":"237","DOI":"10.1097\/NRL.0000000000000626","article-title":"Stroke-associated pneumonia and the brain-gut-lung Axis: a systematic literature review","volume":"30","author":"Bai","year":"2025","journal-title":"Neurologist"},{"key":"ref20","doi-asserted-by":"publisher","first-page":"1736272","DOI":"10.3389\/fmed.2025.1736272","article-title":"Multimodal artificial intelligence in medicine: a task-oriented framework for clinical translation","volume":"12","author":"Zhang","year":"2025","journal-title":"Front Med"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1136\/bmjresp-2023-002249","article-title":"Multimodal modeling with low-dose CT and clinical information for diagnostic artificial intelligence on mediastinal tumors: a preliminary study","volume":"11","author":"Yamada","year":"2024","journal-title":"BMJ Open Respir Res"},{"key":"ref22","doi-asserted-by":"publisher","first-page":"723","DOI":"10.1038\/s43018-022-00388-9","article-title":"Multimodal data integration using machine learning improves risk stratification of high-grade serous ovarian cancer","volume":"3","author":"Boehm","year":"2022","journal-title":"Nat Cancer"},{"key":"ref23","doi-asserted-by":"publisher","first-page":"2666","DOI":"10.1161\/STROKEAHA.118.021981","article-title":"Neuroimaging determinants of Poststroke cognitive performance","volume":"49","author":"Puy","year":"2018","journal-title":"Stroke"},{"key":"ref24","doi-asserted-by":"publisher","first-page":"104482","DOI":"10.1016\/j.jbi.2023.104482","article-title":"A scoping review on multimodal deep learning in biomedical images and texts","volume":"146","author":"Sun","year":"2023","journal-title":"J Biomed Inform"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.3390\/bioengineering12090954","article-title":"Eye tracking-enhanced deep learning for medical image analysis: a systematic review on data efficiency, interpretability, and multimodal integration","volume":"12","author":"Duan","year":"2025","journal-title":"Bioengineering"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1162\/imag_a_00214","article-title":"Applications of interpretable deep learning in neuroimaging: a comprehensive review","volume":"2","author":"Munroe","year":"2024","journal-title":"Imaging Neurosci"},{"key":"ref27","doi-asserted-by":"publisher","first-page":"820","DOI":"10.1109\/JPROC.2021.3054390","article-title":"A review of deep learning in medical imaging: imaging traits, technology trends, case studies with progress highlights, and future promises","volume":"109","author":"Zhou","year":"2021","journal-title":"Proc IEEE Inst Electr Electron Eng"},{"key":"ref28","doi-asserted-by":"publisher","first-page":"104405","DOI":"10.1016\/j.medengphy.2025.104405","article-title":"Explainable deep learning framework for brain tumor detection: integrating LIME, grad-CAM, and SHAP for enhanced accuracy","volume":"144","author":"Akgundogdu","year":"2025","journal-title":"Med Eng Phys"},{"key":"ref29","doi-asserted-by":"publisher","first-page":"33","DOI":"10.1186\/s40708-025-00279-6","article-title":"Boosting brain tumor detection with an optimized ResNet and explainability via grad-CAM and LIME","volume":"12","author":"Afnaan","year":"2025","journal-title":"Brain Inform"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"3117","DOI":"10.2147\/IJGM.S524450","article-title":"Predicting stroke-associated pneumonia in acute ischemic stroke: a machine learning model development and validation study with CBC-derived inflammatory indices","volume":"18","author":"Xie","year":"2025","journal-title":"Int J Gen Med"},{"key":"ref31","doi-asserted-by":"publisher","first-page":"118","DOI":"10.1186\/s12880-024-01285-6","article-title":"Refining neural network algorithms for accurate brain tumor classification in MRI imagery","volume":"24","author":"Alshuhail","year":"2024","journal-title":"BMC Med Imaging"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"2335","DOI":"10.1161\/STROKEAHA.115.009617","article-title":"Diagnosis of stroke-associated pneumonia: recommendations from the pneumonia in stroke consensus group","volume":"46","author":"Smith","year":"2015","journal-title":"Stroke"},{"key":"ref33","doi-asserted-by":"publisher","first-page":"1006","DOI":"10.1038\/mt.2011.95","article-title":"AAV9: over the fence and into the woods","volume":"19","author":"Forsayeth","year":"2011","journal-title":"Mol Ther"},{"key":"ref34","doi-asserted-by":"publisher","first-page":"1077234","DOI":"10.3389\/fnhum.2023.1077234","article-title":"Neural basis of dysphagia in stroke: a systematic review and meta-analysis","volume":"17","author":"Qin","year":"2023","journal-title":"Front Hum Neurosci"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"101685","DOI":"10.1016\/j.nicl.2019.101685","article-title":"Mapping acute lesion locations to physiological swallow impairments after stroke","volume":"22","author":"Wilmskoetter","year":"2019","journal-title":"Neuroimage Clin"},{"key":"ref36","doi-asserted-by":"publisher","first-page":"74","DOI":"10.1016\/j.jstrokecerebrovasdis.2015.08.037","article-title":"Dysphagia post subcortical and Supratentorial stroke","volume":"25","author":"Wan","year":"2016","journal-title":"J Stroke Cerebrovasc Dis"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"3552","DOI":"10.1177\/0300060518775290","article-title":"Swallowing outcomes in patients with subcortical stroke associated with lesions of the caudate nucleus and insula","volume":"46","author":"Im","year":"2018","journal-title":"J Int Med Res"},{"key":"ref38","doi-asserted-by":"publisher","first-page":"116189","DOI":"10.1016\/j.neuroimage.2019.116189","article-title":"Automated anatomical labelling atlas 3","volume":"206","author":"Rolls","year":"2020","journal-title":"NeuroImage"},{"key":"ref39","doi-asserted-by":"publisher","first-page":"565","DOI":"10.1177\/0272989X06295361","article-title":"Decision curve analysis: a novel method for evaluating prediction models","volume":"26","author":"Vickers","year":"2006","journal-title":"Med Decis Mak"},{"key":"ref40","doi-asserted-by":"publisher","first-page":"1513732","DOI":"10.3389\/fneur.2025.1513732","article-title":"Interpretable machine learning for early predicting the risk of ventilator-associated pneumonia in ischemic stroke patients in the intensive care unit","volume":"16","author":"Cao","year":"2025","journal-title":"Front Neurol"},{"key":"ref41","doi-asserted-by":"publisher","first-page":"950662","DOI":"10.3389\/fneur.2022.950662","article-title":"Patient's care bundle benefits to prevent stroke associated pneumonia: a meta-analysis with trial sequential analysis","volume":"13","author":"Guo","year":"2022","journal-title":"Front Neurol"},{"key":"ref42","doi-asserted-by":"publisher","first-page":"297","DOI":"10.1186\/s12883-022-02826-8","article-title":"Reducing the incidence of stroke-associated pneumonia: an evidence-based practice","volume":"22","author":"Liu","year":"2022","journal-title":"BMC Neurol"},{"key":"ref43","doi-asserted-by":"publisher","first-page":"114","DOI":"10.1134\/S0362119715050023","article-title":"The role of insular cortex in autonomic control","volume":"41","author":"Aleksandrov","year":"2015","journal-title":"Fiziol Cheloveka"},{"key":"ref44","doi-asserted-by":"publisher","first-page":"549","DOI":"10.1007\/s10286-024-01066-9","article-title":"The insular cortex, autonomic asymmetry and cardiovascular control: looking at the right side of stroke","volume":"34","author":"Fontes","year":"2024","journal-title":"Clin Auton Res"},{"key":"ref45","doi-asserted-by":"publisher","first-page":"326","DOI":"10.1002\/iid3.277","article-title":"Ischemic stroke alters immune cell niche and chemokine profile in mice independent of spontaneous bacterial infection","volume":"7","author":"Farris","year":"2019","journal-title":"Immun Inflamm Dis"},{"key":"ref46","doi-asserted-by":"publisher","first-page":"991","DOI":"10.1007\/s10571-022-01238-z","article-title":"Functional two-way crosstalk between brain and lung: the brain-lung Axis","volume":"43","author":"Li","year":"2023","journal-title":"Cell Mol Neurobiol"},{"key":"ref47","doi-asserted-by":"publisher","first-page":"519","DOI":"10.4103\/1673-5374.380869","article-title":"Immune regulation of the gut-brain axis and lung-brain axis involved in ischemic stroke","volume":"19","author":"Xie","year":"2024","journal-title":"Neural Regen Res"},{"key":"ref48","doi-asserted-by":"publisher","first-page":"30","DOI":"10.1186\/s12883-023-03488-w","article-title":"Predictive value of cytokines combined with human neutrophil lipocalinin acute ischemic stroke-associated pneumonia","volume":"24","author":"Zhang","year":"2024","journal-title":"BMC Neurol"},{"key":"ref49","doi-asserted-by":"publisher","first-page":"5193","DOI":"10.1016\/j.acra.2024.06.025","article-title":"DeepSAP: a novel brain image-based deep learning model for predicting stroke-associated pneumonia from spontaneous intracerebral hemorrhage","volume":"31","author":"Qiao","year":"2024","journal-title":"Acad Radiol"}],"container-title":["Frontiers in Neurology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fneur.2026.1917270\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T15:36:27Z","timestamp":1789400187000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fneur.2026.1917270\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,14]]},"references-count":49,"alternative-id":["10.3389\/fneur.2026.1917270"],"URL":"https:\/\/doi.org\/10.3389\/fneur.2026.1917270","relation":{},"ISSN":["1664-2295"],"issn-type":[{"value":"1664-2295","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,14]]},"article-number":"1917270"}}